JP2001184082A - Voice processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an easy-to-use voice processor which can accurately and speedily send more necessary information other than an alarm and a warning. SOLUTION: When a word is recognized, weighting values Wa and Wx showing the importance of the word and binary values of excitation time values Ta to Tx set corresponding to the importance of the word are computed and when a computed value T reaches plural thresholds Vths, messages of alarms and warnings corresponding to a specific state are generated, so that a communication means transmits the specific state to other places together with additional information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a terminal device having a security function for merchandise sales, an emergency nurse call, and the recognition of words such as utterances, conversations, and words in portable information terminals used by mobile phones and security personnel. The present invention relates to an audio processing device.

[0002]

2. Description of the Related Art In recent years, the number of stores operating 24 hours a day has increased, and as a result, crimes for stealing cash from electronic cash registers and POS terminal devices have increased. In order to cope with such a situation, various types of merchandise sales data registration devices having a security function have been proposed.

For example, Japanese Unexamined Patent Application Publication No. 10-269455 discloses that a POS terminal device includes an abnormal state detecting means for detecting an abnormal state, a report destination telephone number and a report according to the abnormal state detected by the abnormal state detecting means. A report processing means for selecting the content and reporting via a public communication line is provided, and the abnormal state detecting means is, for example, for detecting a drawer abnormal state in which the drawer is left open for a long time or not for normal operation of the keyboard. Information about the detection of abnormal operation status etc. is disclosed, and if the store manager or employee is forced by the criminal to open the drawer, the risk to the store manager or employee is minimized. In addition, a commodity sales data registration device capable of giving a quick alarm or notification is shown.

[0004]

However, this publication only relates to an audio device for reporting the danger of a terminal device having a product sales security function.
There is no function to recognize voices, conversations and words under specific circumstances with voice input means, and to generate fine-grained alarms and warnings according to the importance of words recognized by voice recognition means In addition, there is a shortage of information in situations where multiple alerts or warnings according to the situation need to be sensed and prompt responses must be taken, or promptness may be sacrificed when trying to increase the amount of information. Was. Also, in an emergency, the recognition of words may not be sufficient because the occurrence of words becomes unclear. Further, when there is no response from the receiving side, it is necessary to repeatedly input to the voice recognition means. Further, when a message such as an alarm or a warning is generated by mistake, it is necessary to input a wrong message to the voice recognition means.

For example, there is a strong need for a voice processing apparatus for recognizing utterances, conversations, and words in a portable information terminal used by a nurse call, a mobile phone, a security guard, or the like, so that both the amount of information and the speed of the situation are compatible. in spite of,
The system described in Japanese Patent Application Laid-Open No. H10-269455 cannot satisfy the above-mentioned requirements. In particular, 5W such as ID information, time, place, etc. relating to a person
Information about 1H is always a very important item.
The transmission of multiple pieces of additional information, such as time, place, facial shape, age, gender, address, telephone number, and e-mail address of a specific person, along with alarms and warnings, may not be possible in emergency situations. It is extremely difficult to communicate accurately and quickly to the side.

Therefore, the present voice recognition device achieves a balance between the amount of information relating to the situation and the speed thereof, and detects only a minimum number of speech signals, simultaneously with a plurality of alarms and warnings, and simultaneously issues both an alarm and a notification. Provided is a voice recognition device capable of transmitting a lot of information.

[0007]

According to a first aspect of the present invention, there is provided a voice processing apparatus having voice recognition means, counting means, and threshold value determination means, wherein a word is recognized by the voice recognition means. Then, the counting unit calculates two values of a weight value indicating the importance of the word and an excitation time value set according to the importance of the word, and the calculated value reaches the set threshold by the threshold determination unit. Is determined,
It is characterized by generating a message such as an alarm or a warning according to a specific situation.

According to a second aspect of the present invention, there is provided an audio processing device comprising:
It is characterized by having communication means for transmitting a message such as an alarm or a warning according to a specific situation.

According to a third aspect of the present invention, there is provided an audio processing apparatus comprising:
The speech recognition means is for recognizing a word as a word registered in advance based on a similar word dictionary.

[0010] According to a fourth aspect of the present invention, there is provided an audio processing apparatus comprising:
It is characterized by having repeat means for repeatedly generating a message such as an alarm or a warning.

[0011] According to a fifth aspect of the present invention, there is provided an audio processing apparatus comprising:
It is characterized by having a message canceling means for canceling a message once generated such as an alarm or warning.

According to a sixth aspect of the present invention, there is provided an audio processing apparatus comprising:
In the counting or threshold determination means, a memory for storing predetermined ID information is provided, and the information stored in the memory is
It is characterized by being transmitted in addition to messages such as alarms and warnings.

The operation of the above configuration will be described below.

According to the processing apparatus of the present invention, when words such as utterances, conversations, and words in a specific situation are recognized by the voice input means, weighting values Wa to Wx indicating the importance of the words and the importance of the words are obtained. Calculates two values of the detection time values Ta to Tx set according to the degree, and generates a plurality of messages such as a plurality of alarms and warnings according to a specific situation when the calculated value T reaches a plurality of thresholds Vth set. Since it is possible, it is possible to process a new original word as a detailed warning or warning as an optimum warning according to the situation and quickly issue it.

The important additional information of 5W1H other than the alarm can be simultaneously transmitted to other places by the communication means.
In the event of an emergency, the occurrence of words becomes unclear, and even when words are not sufficiently recognized, it can be corrected to a message such as an appropriate warning or warning by using a similar term dictionary. Further, when there is no response from the receiving side, it is not necessary to repeatedly input to the voice recognition means, and transmission can be performed without loss by the repeat function.

Further, even if a message such as an alarm or a warning is generated by mistake, it is not necessary to input a false report to the voice recognition means again. Can be canceled. For example, a voice processing device in a personal information terminal or the like used by a nurse call, a mobile phone, a security guard, or the like, selects an optimal alarm or report with a minimum voice signal, and selects a 5W1H that does not require any other selection. It can be transmitted together with the related additional information.

[0017]

Embodiments of the present invention will be described below.

(Embodiment 1) In this embodiment, an embodiment corresponding to claim 1 will be described. FIG. 1 has a recognition unit 1, a counting and judging unit 3, a threshold judging unit 4 and a communication unit 5 for recognizing utterances, conversations and words in a specific situation. Recognizes words, accumulates the importance of words recognized by voice recognition means and counting means within a specific time, and corresponds to each of the plurality of thresholds when the plurality of thresholds set by threshold determination means are reached Audio processor to generate different alarms and warnings
It is a conceptual diagram of 1. Here, 3 is a counting and determining means, 4 is a word table, 5 is a weight value (Wa to Wx), 6 is a threshold value (Vth), and 9 is an excitation time. In FIG. 1, utterances, conversations and words in a specific situation are recognized, the importance of the recognized words is accumulated within a specific time, and an alarm or a warning is transmitted to another place by a communication means when the set threshold is reached. Processing device.

A voice processing device installed in a convenience store where security must be ensured, a nurse call from a hospitalized hospital room, a specific situation such as nursing care for a solitary elderly person, etc. Is recognized by the voice input means 1, then converted into a digital signal by the voice recognition circuit 2, and is counted and counted by the counting and determining means 3 with the table 4 of words matching the recognition content and the digital signal from the voice recognition circuit 2. I do. A weighting value 5 is assigned in advance to the table 4 of words that match the recognition content according to the importance of the words. Two values of weighting values Wa to Wx indicating the importance of the word and detection time values Ta to Tx set according to the importance of the word are calculated, and the calculated value T is compared with a plurality of set thresholds Vth. When a plurality of set thresholds 6 (Vth) are reached, an optimal message such as an alarm or a warning is generated according to a specific situation.

The importance of the words recognized by the voice recognition means and the counting means is accumulated within a specific time, and when a plurality of thresholds set by the threshold determination means are reached, different alarms corresponding to the plurality of thresholds are obtained. And a threshold value 6 (Vth) of the voice processing apparatus that generates a warning and a message such as an optimum warning or warning according to a specific situation. In this way, an optimum message according to the situation can be selected from a plurality of messages such as alarms and warnings.

(Embodiment 2) This embodiment 2 is directed to a second embodiment corresponding to claim 2. Figure 3 recognizes utterances, conversations and words in a specific situation, accumulates the importance of the recognized words within a specific time, and transmits an alarm or warning to another place by communication means when a set threshold is reached. FIG. 2 is a conceptual diagram of a processing device that performs processing. Speech, conversation, and words are input from a microphone, etc. in a voice processing device installed in a convenience store where security must be ensured, a nurse call from a hospitalized hospital room, a specific situation such as nursing aged alone, etc. The speech signal is recognized by the means 1 and then converted into a digital signal by the speech recognition circuit 2, and the counting and judging means 3 compares the digital table from the speech recognition circuit 2 with the table 4 of words matching the recognition contents. In table 4 of words that match the recognition content,
A weight value 5 is assigned in advance according to the importance of the word. The content of the warning is determined by adding the values Wa to Wx of the weighting value 5 indicating the importance of the word, and comparing the added cumulative value T with a plurality of set thresholds Vth6. When a plurality of set thresholds 6 (Vth) are reached, an optimal message such as an alarm or a warning is generated according to a specific situation. 1 is a voice input unit, 2 is a voice recognition circuit, 10 is a message supervisor, 11 is a display unit, 12 is a repeat unit, 13 is a false alarm release unit, and 14 is an ID information storage unit.

The message such as the optimum alarm or warning is transmitted by the communication means 7. What is used as the communication means 7 is
The Internet such as a wired telephone line, a mobile phone, a word processor or a personal computer with a voice recognition function, an in-house LAN, and wired or wireless communication means are used. Messages such as alarms and warnings are transmitted to the optimal destinations such as the police station, fire department, nursing room, nurse station, security staff waiting room, and night shift room.

(Embodiment 3) In Embodiment 3, an embodiment corresponding to claim 3 will be described. FIG. 4 shows specific information registered in a memory in advance without using voice input means, except for an optimal message such as an alarm or warning, which indicates a specific situation selected from a plurality of messages such as alarms or warnings. The speech processing device has a function of translating and recognizing words similar to words registered in advance into unified words based on the similar word dictionary 15 when recognizing words. For example, "Help", "Help?" And "Help?"
here? Represents unrecognizable. The three words "Takete", "Takete" and "Take ??" are translated into recognized words "Takete" as a pre-registered word and recognized.

In an emergency, the utterance may become unclear, or the speed or tone of a word may deviate significantly from the normal state. By adding such a function of the similar word dictionary 15, even if the utterance becomes unclear or the speed or tone of the word shifts, the similar words can be unified to similar words registered in advance by the similar word dictionary. The warning and warning information is accurately transmitted to the information recipients in other places such as police station, fire department, nursing room, nurse's station, security staff waiting room, night shift room, etc. As a result, the information sender is Appropriate measures can be taken.

(Embodiment 4) Embodiment 4 describes an embodiment corresponding to claim 4. FIG. 5 is a schematic diagram of a voice processing apparatus having a function of repeatedly transmitting a plurality of messages such as warnings and warnings. In the event of an emergency, repeated alerts and warnings were accurately transmitted to the information recipients at police stations, fire departments, nursing rooms, nurse stations, security personnel waiting rooms, night duty rooms, etc. Nevertheless, there is a case where there is no response from the receiving side to the transmitting side, which may cause psychological anxiety. As a result of repeating the same voice input each time, psychological fatigue and physical fatigue are further increased. Until the information sender obtains an appropriate response, the repeat means 12 repeatedly generates a message such as an alarm or a warning, thereby avoiding an increase in psychological fatigue and physical fatigue.

The repeat means 12 stores the words once uttered and registered in a memory 17 comprising a flash memory, a micro-cassette tape, an MD, a CD-R, or the like, and causes the repeat means 12 to function, thereby repeatedly issuing an alarm or warning. Since the message can be generated, the sender of the information repeatedly generates messages such as alarms and warnings by using the repeat means until an appropriate response is obtained, thereby increasing psychological fatigue and physical fatigue. Can be avoided.

(Embodiment 5) This embodiment 5 describes an embodiment corresponding to claim 5. FIG. 2 shows an example in which a message such as an optimum alarm or warning that informs a specific situation selected from a plurality of messages such as alarms or warnings is stored in a memory via voice input means. A circuit for giving the function of canceling the once-generated alarm or warning message to the system control CPU by turning on the switch in the means for canceling the once-generated alarm or warning message or the like is provided. This is an example.

If the user notices that a message such as an alarm or a warning has been mistakenly given by mistake, and re-enters the message as a false report via the voice recognition means, the receiving side will determine which message is correct and which message is incorrect. Or was confused. However, by correctly transmitting the correct message from the system control CPU, it is possible to eliminate the embarrassment of the receiving side.

By turning on the false alarm canceling means, it is possible to efficiently cancel messages such as alarms and warnings.
By receiving such information, the receiver can take appropriate action promptly.

(Embodiment 6) This embodiment 6 will describe an embodiment corresponding to claim 6. FIG. 4 shows a specific message registered in advance in the memory 17 without any voice input means, except for an optimal message such as an alarm or warning that informs a specific situation selected from a plurality of messages such as alarms or warnings. Information, such as location, time, ID information given to a specific person such as who is the caller, along with a map, face photo, address, telephone number, e-mail address, etc. FIG. 2 is a diagram illustrating a system configuration for transmission.

In the case of an emergency, alarm and warning information alone is insufficient, and the information recipients are located in other places such as a police station, a fire station, a nursing room, a nurse's station, a security staff waiting room, a night shift room, etc. It is necessary to take appropriate measures by receiving accurate 5W1H information. However, in an emergency, the sender of the information cannot send a detailed message.
As a method for resolving such inconsistency, ID information given to a specific person such as a place, a time, and a caller is alerted together with a map, a face photograph, an address, a telephone number, an e-mail address, and the like. This is a system configuration in which the information is transmitted by a communication unit together with a warning and a warning. By transmitting according to FIG. 4, the receiver can receive a more detailed message. The ID information assigned to a specific person, such as the location, who the caller is, etc., is stored in the ROM in advance for the map, face photograph, address, telephone number, and e-mail address,
Send an alarm. For the time, the information of the clock built in the alarm is transmitted online.

Such information can be obtained in a voice processing apparatus installed in a particular location such as a convenience store where security must be ensured, a nurse call from a hospitalized hospital room, or the care of an elderly person living alone. , It is possible to take appropriate action promptly.

[0033]

As described above, in the processing apparatus of the present invention, a detailed warning or warning corresponding to the importance of the word recognized by the voice recognition means can be issued by an unspecified user such as an elderly person, a sick person or a woman. At the same time, there is an effect that it is possible to easily generate an alarm without learning, and it is also possible to finely extract an optimum alarm according to a situation and quickly issue the alarm. There is an effect that transmission can be quickly performed to another optimum place by various communication means. In the event of an emergency, even when the recognition of words is not sufficient, the similar term dictionary is used to replace the message with an appropriate warning, warning, or the like, so that it is possible to accurately and quickly issue the message.

When there is no response from the receiving side, there is no need to repeatedly input the information to the speech recognition means, and the information is transmitted repeatedly without loss by the repeat function until the appropriate response is obtained by reducing the psychological and physical fatigue of the information sender. The effect that can be performed. When a message such as an alarm or a warning is erroneously reported by mistake, there is no need to input a false report to the voice recognition means again, and the warning or warning message can be efficiently canceled by the false report canceling means.

The voice processing device is provided with a memory for storing ID information assigned to a predetermined specific person,
By transmitting a plurality of pieces of additional information such as the figure and age of a face related to a specific person, age, gender, address, telephone number, and e-mail address stored in the memory with the transmission time and the memory as a warning or warning, a warning or warning can be promptly issued. This has the effect of providing a voice processing device capable of transmitting a lot of information together with the notification.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram showing Embodiment 1 of the present invention.

FIG. 2 shows Embodiment 2, Embodiment 3, Embodiment 4,
33 is a block diagram illustrating a configuration according to Embodiments 5 and 6. FIG.

FIG. 3 is a diagram illustrating a relationship between a weight value and an excitation time according to the first embodiment of the present invention.

FIG. 4 is a diagram showing an example of a detailed configuration of a speech recognition circuit 2 of the present invention.

FIG. 5 is a diagram illustrating a relationship between an accumulated value and time according to the first embodiment of the present invention.

FIG. 6 is a diagram showing a configuration of a message supervisor.

FIG. 7 is a diagram illustrating an example of a display board of a message supervisor.

FIG. 8 is a flowchart showing an input determination process of a conventional voice processing device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Speech input means 2 Speech recognition circuit 3 Counting and judgment means 4 Word table 5 Weight value (Wa-Wx) 6 Threshold value (Vth) 7 Communication means 8 Words 9 Excitation time 10 Message supervisor 11 Display means 12 Repeat means 13 false information canceling means 14 ID information storage unit 15 similar word dictionary 16 alarm bell 17 memory

 ────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Hiroshi Suzuki 22-22 Nagaikecho, Abeno-ku, Osaka-shi, Osaka F-term (reference) 5D015 KK01 LL03 LL10

Claims (6)

[Claims] 1. A speech processing apparatus having speech recognition means, counting means, and threshold value judgment means, wherein when a word is recognized by the speech recognition means, a weighting value indicating importance of the word is given by the counting means. Calculates the two values of the excitation time value set according to the importance of the word, and when it is determined that the calculated value has reached the set threshold value by the threshold value determination means, an alarm or a warning according to a specific situation is issued. An audio processing device for generating a message. 2. The voice processing apparatus according to claim 1, further comprising communication means for transmitting a message such as an alarm or a warning according to a specific situation. 3. The speech processing apparatus according to claim 1, wherein the speech recognition means recognizes the words as words registered in advance based on a similar word dictionary. 4. The apparatus according to claim 1, further comprising a repeat unit for generating a message such as an alarm or a warning repeatedly.
The voice processing device according to 1. 5. The voice processing apparatus according to claim 1, further comprising a message canceling unit for canceling a once-generated message such as an alarm or a warning. 6. A memory for storing predetermined ID information in a counting or threshold value judging means, wherein the information stored in the memory is transmitted in addition to a message such as an alarm or a warning. The audio processing device according to claim 1, which performs the processing.